Gist Abstraction Mechanism Of Natural Scenes

Complex scenes, also called natural scenes, are visual stimuli composed of more than one objects. Investigating into the gist abstraction mechanisms of natural scenes is an effective way to understand human cognitive mechanisms. Currently, there are several theories about the gist abstraction of natural scenes, including objects based theory, object-relations based theory, scene entity theory and so on. All of them are supported by certain experimental evidence. The reason for the differences is that the scene cognitive process is an integrated and dynamic process.We presume that the gist abstraction of natural scenes is based on semantic meaning of component objects and the relations among them. Based on this hypothesis, we designed an interview and two experiments. In the interview, we asked twenty subjects to report gist of a large number of pictures, as to understand how people get natural scene gist. In experiment 1, we manipulated presentation time and details of pictures, as to learn how people use scene information in different stages. In experiment 2, we observed response time and accuracy of searching for a certain target object, as to make clear of how object importance functions in gist abstraction and how scene gist influences object recognition. These are the conclusions:1. During the process of gist abstraction of natural scenes, semantic meaning of objects and relations among them are both used. The extent to which each kind of information is used depends on scene features and scene readers' experience.2. As viewing time increases, gist abstraction of natural scenes tends to use more specific information.3. Objects in a scene compete with each other, so that certain objects get more importance. These objects are dominant in gist content, and can be processed more quickly in gist abstraction.4. People integrate scene information to form gist, and the obtained gist can influence people understand scene in the following time. This cognitive process is dynamic and full of trading offs.